/ic/ - Artwork/Critique

>>6293182
You fags don't even know how to argue your case.
Looking and learning from copyrighted material is allowed for humans, right?
That's exactly what the AI does. There aren't tiny pieces of copyrighted images stored in its data. In some real sense, the weights of the neural network store concepts and ideas and they vary enormously in their scope from say a tiny dust particle or a basic shape to the knowledge about high level things like the human figure, animals and buildings and also in-between stuff, which cannot even be named because we're unaware of it, since we don't know the exact way both the human brain and AI neural networks function.
So really your only hope is to distinguish AIs from a humans, based on some rights the latter have and then you're basically down to a luddite argument.
And it's not even the case that I have no sympathy or understanding why that might be needed, especially given the possibility of future improvements in this technology.
Anyway, let's have a thought experiment. Imagine that we have these AIs trained only on data, which is public domain or the rights to which have been secured for that purpose. I can assure you, that sooner or later such AI will become incredibly good too. What will your argument be then?